2009 TASI Lecture Introduction to Extra

Dimensions
arXiv:1003.1162v1 [hep-ph] 4 Mar 2010

Hsin-Chia Cheng

Department of Physics, University of California

Davis, California 95616, USA

Abstract

We give a brief introduction to theories with extra dimensions. We first introduce

the basic formalism for studying extra-dimensional theories, including the Kaluza-Klein
decomposition and the effective theory for 3-branes. We then focus on two types of
scenarios: large extra dimensions as a solution to the hierarchy problem and TeV1 -
size extra dimensions with Standard Model fields propagating in them. We discuss the
experimental tests and constraints on these scenarios, and also the questions in particle
physics which may be addressed with the help of extra dimensions. This is the write-up
of the lectures given at the 2009 TASI summer school. Other interesting topics such as
warped extra dimensions are covered by other lecturers.

1
1 Introduction
The idea of extra dimensions started from attempts to unify different forces in nature. In
1914 Nordstrom [1] proposed a 5-dimensional (5D) vector theory to simultaneously describe
electromagnetism and a scalar version of gravity. After the discovery of General Relativity,
Kaluza [2] (1919) and Klein [3] (1926) realized that the 5D Einsteins theory with one spatial
dimension compactified on a circle can describe both the 4-dimensional (4D) gravity and
electromagnetism. However, the Kaluza-Klein (KK) theory has many problems and is not
a viable model to describe nature. Higher dimensional theories received renewed interests
in the late 1970s and 1980s because of the developments in supergravity and superstring
theories. The consistency of superstring theory requires extra dimensions. However, the
extra dimensions considered then are extremely small, of the order Planck length Mpl1 , which
is beyond any possible experimental reach.
In the 1990s people began to consider the possibility that some extra dimensions are much
larger than the Planck length:

Antoniadis [4] (1990) proposed TeV1 -size extra dimensions related to supersymmetry
(SUSY) breaking.

Horava and Witten [5, 6] (1996) noticed that an extra dimension (1012 GeV)1 in
M-theory can lower the string scale to the grand unification scale MGUT 1016 GeV
and hence can unify gravity together with other forces at the same scale.

The discovery of D-branes in string theory by Polchinski [7] (1995) provides a natural
setting for different fields living in different number of extra dimensions, e.g., Stan-
dard Model (SM) fields can be described by open strings which are localized on lower-
dimensional D-branes, while gravitons are described by closed strings which propagate
in all dimensions.

The idea of extra dimensions became popular in phenomenology after Arkani-Hamed,

Dimopoulos and Dvali [8] (1998) considered large extra dimensions as a solution to
the hierarchy problem. The way to address the hierarchy problem with large extra
dimensions is described below.

Warped extra dimensions (Randall and Sundrum [9, 10], 1999) and AdS/CFT corre-
spondence (Maldacena [11], 1998) provide new exciting possibilities to understand and
construct models related to the weak scale. They are covered in Tony Gherghettas
lectures [12], so we will focus on flat extra dimensions here.

2
 Figure 1: Force lines from a point mass in compact extra dimensions

Large extra dimensions as a solution to the hierarchy problem: In Standard Model

the electroweak symmetry is broken by the vacuum expectation value (VEV) of a scalar
Higgs field. However, the electroweak scale is unstable under radiative corrections as the
mass-squared of a scalar field receives quadratic contributions from its interactions. The
natural scale to cut off the quadratic contributions is the Planck scale when the quantum
gravity effects become important. Then, the question is why electroweak scale ( 100 1000
GeV) is so much smaller than the 4D Planck scale ( 1019 GeV). One can turn this question
around and ask: why is gravity so weak compared with other interactions in the Standard
Model. One possibility is that the 4D Planck scale may not be a fundamental scale and the
scale of quantum gravity is actually much lower if there exist large extra dimensions [8, 13, 14].
Let us consider Newtons law in 4 + n dimensions:
(4+n)
GN m1 m2 1 m1 m2
F (r) = n+2 . (1.1)
r n+2 Mpl(4+n) r n+2

If n extra dimensions are compact with size L = 2R, then the force lines from a source mass
have to go parallel in extra dimensions when the distance in the usual 3 spatial dimensions is
larger than L (Fig. 1),
1 m1 m2
F (r) n+2 , for r L, (1.2)
Mpl(4+n) r n+2
1 m1 m2
F (r) n+2 , for r L. (1.3)
Mpl(4+n) Ln r 2

Comparing the last expression with the 4D Newtons law,

1 m1 m2
F (r) n+2 , (1.4)
Mpl(4) r2

we have
2 n+2
Mpl(4) Mpl(4+n) Ln = Mpl(4+n)
n+2
Vn , (1.5)

3
 SM

e Graviton
e

e
e

3-brane

Figure 2: The brane world

where Vn is the volume of the compact extra dimensions. If we take the fundamental scale
Mpl(4+n) 1 TeV and the 4D Planck scale Mpl(4) 1019 GeV, then we have (assuming extra
dimensions have the same size)
!1/n
2
Mpl(4)
L n+2 1032/n TeV1 1032/n 1017 cm, (1.6)
Mpl(4+n)

n = 1 L 1015 cm (> 1 AU), obviously ruled out,

n = 2 L 1 mm , allowed in 1998, but current bound L < 200 m
n = 3 L 106 cm .

On the other hand, SM has been well-tested up to a few hundred GeV to TeV, so SM
field cannot propagate in extra dimensions with size R & 1 TeV1 . If there are large extra
dimensions, SM fields have to be localized on a 3-brane as shown in Fig. 2 (with thickness
. 1 TeV1 ). It was surprising that such a scenario is alive and not ruled out experimentally
or observationally. In Sec. 4 we discuss various constraints from high-energy and low-energy
experiments as well as from astrophysics and cosmology. In the next 2 sections we will first
develop the formalism for studying theories with extra dimensions.

4
 Figure 3: One extra dimension compactified on a circle

2 Kaluza-Klein Theory
We first discuss how a higher-dimensional field theory reduces to 4-dimensional after the
extra dimensions are compactified. As an illustration let us consider one extra dimension
compactified on a circle with radius R (Fig. 3). The coordinates are denoted as xM = (x , y),
where M = 0, 1, 2, 3, 5, = 0, 1, 2, 3 and y = x5 is the coordinate in the direction of the extra
dimension. The compactification means that the points y and y + 2R are identified. We
start with the simplest example of a free real scalar field. The action of a free (massless)
scalar field in 5D is
1
Z
S = d5 x M (x , y) M (x , y). (2.7)
2
The scalar field has mass dimension 3/2 in 5D. Because the extra dimension is compactified,
the field value is periodic in y coordinates, (x , y + 2R) = (x , y). We can perform the
Fourier decomposition of along the y direction:

1 X n
(x , y) = (n) (x )ei R y . (2.8)
2R n=

5

Reality of implies (n) = (n) . Plugging the Fourier series expansion into the action,
"
1 X 1
Z
m n
S = d4 x dy (m) (x)ei R y (n) ei R y
2R m,n 2
    #
1 m (m) m n n
i (x)ei R y i (n) (x)ei R y
2 R R
! " #
1 1 mn
Z X Z m+n
= d4 x dy ei R y (m) (x) (n) (x) + 2 (m) (x)(n) (x)
m,n
2R 2 R
" #
4 1 n2 (n) (n)
Z X
(n) (n)
= dx 2
2 n R
(  )
2
1 n
Z X
= d4 x (0) (0) + (n) (n) 2 (n) (n) . (2.9)
2 n=1
R
We can see from the 4D point of view that the action describes an (infinite) series of particles
(Kaluza-Klein tower) with masses m(n) = n/R. If the field (x , y) has a 5D mass m0 ,
then the 4D Kaluza-Klein particles will have masses, m2(n) = m20 + n2 /R2 . It is also easy to
generalize it to higher dimensions compactified on a torus. In this case the masses of the
Kaluza-Klein states are given by
n25 n26
m2n5 ,n6 ,... = m20 + + + , (2.10)
R52 R62
where R5 , R6 , . . . are the radii of the corresponding compact dimensions.
The next example is a gauge field in 5D, AM (x , y). We can similarly perform a Fourier
decomposition along the compact dimension,
1 X (n) i n y
AM (x , y) = AM (x )e R . (2.11)
2R n
The derivative along the extra dimension can be replaced by 5 i(n/R) under the Fourier
decomposition. The action becomes
 
1
Z
4 MN
S = d x dy FM N F
4
 
1 1
Z
4
= d x dy F F + ( A5 5 A )( A5 5 A )
4 2
"
Z X 1
= d4 x (n) (n)
F F
n
4
  #
1 (n) n (n) n
+ A5 + i A(n) A5 i A(n) . (2.12)
2 R R

6
We can perform a gauge transformation to make A5 constant along the extra dimension to
remove the mixed terms,
1 (n)
A(n) A(n) i A5 ,
n/R
(n)
A5 0, for n 6= 0. (2.13)

In this gauge,
( 
1 (0) (0) 1
Z
4 (0) (0)
S = dx F F + A5 A5
4 2
 2
)
X 1 (n) (n) 1 n (n) (n)
+ 2 F F + A A . (2.14)
n1
4 2 R2

We can see that zero modes contain a 4D gauge field and a real scalar. (For a non-Abelian
(n)
gauge group the scalar would be in the adjoint representation.) For nonzero modes, A5
(n)
is eaten and becomes the longitudinal mode of the corresponding massive vector field A .
There is no scalar mode left for nonzero KK levels.
To match the 5D gauge coupling and the 4D gauge coupling we can examine the 5D
covariant derivative DM = M + ig5 AM . Because AM has mass dimension 3/2 in 5D, the 5D
gauge coupling g5 has mass dimension 1/2. Expanding A into its KK levels,
1
D = + ig5 A = + ig5 A(0)
+ , (2.15)
2R
we see that the 4D gauge coupling is given by
g5
g4 = . (2.16)
2R
Note that the 5D (also higher dimensional) gauge coupling has a negative mass dimension,
so the 5D gauge theory is non-renormalizable. It becomes strongly interacting at energy
scale E 1/g52 = 1/(2Rg42). Thus it can only be treated as a low energy effective theory
with a cutoff 4/g52 . From the 4D point of view, the strong interaction comes from
the enhancement of the number of KK modes accessible at the energy scale. The effective

coupling is NKK g4 .
More generally, if we start with a 4 + n dimensional gauge theory with n dimensions
compactified on a torus, the zero modes will contain a 4D gauge field together with n adjoint
scalars, and each nonzero KK level will have a 4D massive vector field and (n 1) massive
adjoint scalars.

7
 Next we consider the gravitational field in D = 4 + n dimensions. Gravitational field is
described by a symmetric metric tensor gM N = M N +hM N . There are D(D+1)/2 independent
components of a symmetric tensor in D dimensions. Many degrees of freedom can be removed
by the D-dimensional general coordinate transformation hM N hM N + M N + N M . We
1
can impose D conditions to fix the gauge, e.g., harmonic gauge, M hM M
N = 2 N hM . However,
gauge transformations satisfying M = 0 are still allowed. Another D conditions can be
imposed. The number of independent degree of freedom becomes

D(D + 1) D(D 3)
2D = . (2.17)
2 2
For D = 4, 5, 6, they are listed below:

Dimension(D) Number of degree of freedom

4 2
5 5
6 9

On the other hand, a 4D massive spin-2 field has 5 polarizations. A 5D graviton with one
spatial dimension compactified decomposes into

hM N h h5 h55 . (2.18)

Zero modes consist of a 4D graviton, a massless vector, and a real scalar. For nonzero modes,
(n) (n) (n)
h5 and h55 are eaten by h to form massive spin-2 fields.
The generalization to 4 + n dimensions is straightforward. The zero modes contain one 4D
graviton, n massless vectors, and n(n + 1)/2 scalars. For nonzero modes, each KK level has
one massive spin-2 tensor, (n 1) massive vectors, and n(n + 1)/2 1 (n 1) = n(n 1)/2
massive scalars. One can also deduce the relation between the (reduced) Planck scales in 4
dimensions and higher dimensions,
2+n
M 4+n
Z q
4+n
S = d x |g(4+n) | R(4+n)
2
2+n
M 4+n
Z
n
d4 x g(4) R(4) +
p
= (2R)
2
2 Z
M4
d4 x g(4) R(4) +
p
= (2.19)
2
. (2.20)

8
We obtain
2 2+n 2+n
M 4 = (2R)n M 4+n = Vn M 4+n , (2.21)
where the 4D reduced Planck scale
1 Mpl(4)
M4 = = = 2.4 1018 GeV. (2.22)
8GN 8
The decomposition of a higher-dimensional graviton is summarized below [15, 16].
4D graviton and its KK modes: they are at the upper left 4 4 corner of the
(4 + n) (4 + n) matrix for the higher-dimensional graviton,
~
!
Gk
(2.23)

These modes are labeled by the n-dimensional vector ~k which corresponds to the KK numbers
along the various extra dimensions.
4D vectors and their KK modes: they live at the off-diagonal blocks of the higher-
dimensional graviton matrix,
~
!
Vjk
~ (2.24)
Vjk

For massive modes, they satisfy an additional constraint

~
k j Vjk = 0, (2.25)

so there are only n 1 independent vectors.

4D scalars and their KK modes: the lower right n n block of the graviton matrix
corresponds to 4D scalar fields:
!
~ (2.26)
Sijk

The massive modes satisfy the additional constraint

~
k j Sjk
k
= 0, (2.27)

It is also convenient to separate out the radion as a special field which is represented by the
~
trace hkj
j . The other scalars then satisfy the traceless condition

~
Sjkj = 0. (2.28)

9
 The explicit expressions for
q the canonically normalized 4D fields are given in unitary gauge
3(n1)
by (using the notation = n+2
):

~ 1 ~kj
radion Hk = h
j
~ ~ ki kj ~k
scalars Sijk = hkij (ij + )H
n1 k 2
~ i ~
vectors Vjk = hkj
2
~ ~ ~k
gravitons Gk = hk + ( + )H . (2.29)
3 k 2
The equation of motion in the presence of sources is given for the above fields by
~k 1 h
i
G MP l
T + ( + k2 )T /3
V ~k
0
( + k 2 ) j

~k =
. (2.30)
Sij 0


~k
H
T
3MP l

We see that only the 4D graviton, the radion and their KK modes couple to the brane
sources. Other fields do not couple to the matter fields on the brane directly and hence are
not important for processes involving the brane matter fields.

3 Effective Field Theory for a Three-brane

In the large extra dimension scenario, the SM fields have to be localized on a 3-brane. To
describe the interactions with the bulk, we need to develop an effective field theory for a
3-brane. We follow the work by Sundrum [17]. A 3-brane breaks the higher-dimensional
space-time symmetry either spontaneously or explicitly. In the case of spontaneous symmetry
breaking, the 3-brane can fluctuate and there are Nambu-Goldstone bosons associated with
the symmetry breaking. Examples are domain walls and D-branes in string theories. In the
case that the 3-brane does no fluctuate such as the orbifold fixed points or orientifolds in
string theory, the extra space-time symmetry is explicitly broken and there is no associated
Nambu-Goldstone mode. Here we will discuss the spontaneous symmetry breaking case.
We assume that the 3-brane is flat (= R4 ) and the extra dimensions are compactified on a
torus T n . Coordinates in the bulk are denoted as X M , M = 0, 1, 2, . . . , 3 + n, coordinates on
the brane are denoted as x , = 0, 1, 2, 3, and the coordinates along the extra dimensions are
X m , m = 4, 5, . . . , 3 + n. The metric in 4 + n dimensions is GM N (X). The bulk coordinates

10
describing the position occupied by a point x on the 3-brane are denoted as Y M (x) and they
are dynamical fields. SM fields are functions of x, (x), A (x), (x).
The effective field theory describes small fluctuations around the vacuum state. In the
vacuum

GM N (X) = M N (using the mostly negative signature +, , , ),

Y M (x) = M x (a simple gauge choice). (3.31)

The bulk action is given by

2
!
M 4+n
Z p
Sbulk = d4+n x |G| R(4+n) . (3.32)
2

To write down an effective action for the brane localized fields we need the induced metric on
the brane,
Y M Y N
ds2 = GM N dY M (x)dY N (x) = GM N dx dx
x x
= g dx dx . (3.33)

We see that the induced metric is given by

Y M Y N
g = GM N (3.34)
x x
and in the vacuum state g = .
The action is invariant under the general coordinate transformations of the bulk coordi-
nates X M and also the general coordinate transformations of the brane coordinates x , so we
need to contract the indices M and separately. The brane action can be written as
(
M42 1
Z p
4
Sbrane = d x |g| R(4) f 4 + g D D V ()
2 2
)

g g
F F + , (3.35)
4

where f 4 is the brane tension, and the 4D Planck scale receives contributions from both the
2+n
bulk term M 4+n Vn and the brane term M42 . Assuming that the brane tension f M 4+n ,
then we can ignore the back reaction on gravity.
We can use the 4D reparametrization invariance to gauge fix

Y (x) = x , = 0, 1, 2, 3, (3.36)

11
then only Y m (x), m = 4, 5, n + 3 are physical degrees of freedom. Their kinetic terms can
be derived by expanding the tension term of the brane action,
Z p
S = d4 x |g|[f 4 + ],

g = GM N Y M Y N = + Y m Ym + ,
det g = 1 Y m Ym + ,
p 1
|g| = 1 + Y m Ym + ,
2 
1
Z
4 4 m m
S = d x(f ) 1 + Y uYm +
2
4
 
f
Z
4 4
= d x (f ) + Ym Ym + (Y m = Ym ). (3.37)
2

Canonically normalized fields are given by Zm = f 2 Ym . A positive tension (f 4 > 0) implies

a positive kinetic term. On the other hand, if the brane tension is negative, then the kinetic
term is negative and Ym s are ghost field. The system is unstable and the brane would
like to crumble. Negative tension can therefore only occur with explicit breaking of higher-
dimensional space-time symmetry where there is no Nambu-Goldstone bosons Ym .
We can also derive the couplings of the localized SM fields to bulk gravitons. The stress-
energy tensor of the SM fields is given by
2 S
T = p , (3.38)
|g| g

and the interaction is Z p 1

Sint d4 x |g| T g . (3.39)
2
Expanding g we have

g = GM N Y M Y N
= (M N + 4+n HM N )(M + )(N + )
2
= + H + . (3.40)
M 4+n

For extra dimensions compactified on a torus, HM N (X L ) is periodic for L = 4, 5, 3 + n,

(~k)
X X hM N i ~k~y
HM N (x, y) = e R. (3.41)
k = k =
Vn
1 n

12
We can choose ~y = 0 for the 3-brane location, then

X h(~k)
HM N (x, 0) = M N . (3.42)
~
Vn
k

The interactions between the SM fields and KK gravitons are given by

Z p 1 X 2 ~
Sint d4 x |g| T h(k)
2 M 4+n Vn
~k
1 X (~k)
Z p
= d4 x |g| T h . (3.43)
M4 ~ k

4 Experimental Constraints and Tests of Large Extra

Dimensions
Large extra dimensions may appear to be a radical proposal to address the hierarchy problem.
Whenever some possible new physics is proposed, we need to ask whether it is consistent
with the current experimental constraints and how we can test it experimentally. Some
characteristics of large extra dimensions are crucial for experimental constraints and tests:

TeV cutoff: Precision electroweak tests and high energy collisions are sensitive to higher-
dimensional operators suppressed by the TeV scale.

Light degrees of freedom: They can appear as missing energies at colliders and rare
decays of unstable particles. They also affect astrophysics (e.g., star cooling) and cos-
mology (e.g., expansion rate of the universe).

Long-lived KK gravitons: They can affect astrophysics (diffuse -ray background from
late decays of long-lved particles) and cosmology (over-closure of the universe).

We will discuss various experimental bounds and possible tests in the following subsections.
We will only give a flavor of what types of constraints are relevant and make simple estimates.
For more detailed discussion please consult the original papers in the literature.

4.1 Laboratory bounds on long-range forces

At distances shorter than the compactification radius of extra dimensions, the gravitational
force will be modified from the usual inverse-square law. The experimental tests of the

13
Figure 4: Laboratory bounds on deviations of the gravitational inverse-square law, taken from
D. J. Kapner et al [18].

gravitational force are usually parametrized by the modified potential,

m1 m2
1 + er/ ,


V (r) = GN (4.44)
r
where is the distance where the modification occurs and is given by the inverse mass of
the new light particle which mediates the new force, and represents the strength of the
new force relative to the gravitational force. For the large extra dimension scenario, is
the inverse mass of the first KK graviton, = (m(1) )1 = R. and is the number of the
first KK modes (e.g., = 4 for 2 extra dimensions on a torus). The bounds from various
experiments are shown in Fig. 4. For 2 extra dimensions of the same size, the current bound
is R < 37m [18]. Using the relation between the reduced Planck scales in 4 dimensions and
in 6 dimensions,
2 4
M 4 = (2R)2 M 6 , (4.45)

14
we obtain a bound on the 6-dimensional Planck scale,
s
M4
M6 = > 1.4 TeV. (4.46)
2R

As we will see later, there are more stringent bounds from astrophysics and cosmology. For
more than 2 extra dimensions, the laboratory constraints on long-range forces do not place
any significant bounds on the higher-dimensional Planck scale.

4.2 Particle physics constraints

The constraints on large extra dimensions from particle physics can be divided into several
categories.

4.2.1 Higher dimensional operators

If the fundamental Planck scale as a cutoff of the SM physics is at TeV, we expect that
there will be higher-dimensional operators suppressed by the cutoff scale,
ci 4+m
L O with M4+n TeV. (4.47)
m i
1. Some operators are strongly constrained, , e.g.,

proton decay & 1015 GeV,

Majorana neutrino 1015 GeV,
FCNC,CP
6 & 100 1000 TeV. (4.48)

They appear to be dangerous (but one faces the same problem in any extension of
the SM at the TeV scale). However, these operators violate (approximate) symmetries
of SM, so one can imagine ways to suppress them, such as gauging baryon number
symmetry and imposing flavor symmetries etc.

2. For operators which respect symmetries of SM, the constraints come from the elec-
troweak precision tests. In general & hundreds GeV 10 TeV. They are reviewed in
Prof. Skibas lectures [19]. It causes a little hierarchy problem if M4+n 1 TeV.

3. Some operators can be induced by KK graviton exchanges [15, 16, 20, 21]. For example
e+ e e+ e through KK gravitons (Fig. 5) induces an operator,

15
 e+ e+
KK graviton

e- e-

Figure 5: A four fermion interaction induced by KK graviton exchanges.

KK graviton

Figure 6: A Feynman diagram which contribute to muon g 2 with KK graviton loop.

X E2 1
2
LO c 2 , (4.49)
~k M4 |~kR1 |2

where we have omitted the tensor structure. The sum is logarithmic divergent for n = 2
and power divergent for n > 2, so it needs to be cut off,

E 2 n2
2
LO 2+n . (4.50)
M 4+n

If we take the cutoff M 4+n , then M 4+n & 1 TeV from current experimental con-
straints. Another example is muon g 2 where the KK graviton loop gives rise to a
divergent contribution which needs to be cut off (Fig. 6) [22], It gives similar constraint.

4.2.2 Real emission of KK gravitons

This set of experimental tests relies on the existence of light KK gravitons. It is less sensitive
to the cutoff physics and hence more robust. The mass splitting of KK modes in large extra
dimensions is
  n2   n+2
1 M 4+n M 4+n n
12n31
m M 4+n = 10 n eV. (4.51)
R M4 TeV

16
 W
KK graviton

s d
u, c, t

Figure 7: A diagram which contribute to K + KK graviton decay.

It is extremely small in particle physics scale and the KK spectrum can be treated as contin-
uum for any practical purpose. We can estimate the number of KK modes with momentum
in extra dimensions between k and k + dk,

dN = Sn1 k n1 dk
= Sn1 mn1 Rn1 dm R
2
Sn1 M 4 1
= m dm, (4.52)
(2)n M n+2
4+n
n1
where Sn1 = (2) /( n1
2
2
) is the surface area of an n 1 dimensional sphere with unit
radius, and m = |~k|/R. The number of KK modes accessible at energy E is
2  n
M4 E
N(E) 2 . (4.53)
M 4+n M 4+n
We see that there are more accessible KK modes at higher energies, which implies stronger
constraints will come from processes with higher energies. On the other hand, for E < M 4+n ,
there are fewer KK modes for larger n. It implies weaker constraints for larger n from the
same process.

1. Rare decays to KK gravitons: An example is K + KK graviton (Fig. 7) [14]. In

the 4D picture, decay width to a single KK graviton is
1 m5K m2K
1 . (4.54)
16 m4W M 2
4

The number of KK gravitons with mass less than mK is (mK R)n , so the total decay
width of K to + KK graviton is
1 m5K m2K n n
K+KK graviton 4 2 mK R
16 mW M
4
5
 n+2
1 mK mK
. (4.55)
16 m4W M 4+n

17
 - , Z, g - , Z, g - , Z, g
f f f

f GKK f GKK f GKK

Figure 8: Feynman diagrams for the KK graviton production.

We see that the result is proportional to the square of the higher dimensional gravi-
n+2
ton coupling M 4+n2 as expected. One can obtain the same estimate from a higher
dimensional picture. The current experimental bound on the rare K decay is

B(K + X) < 1010 , (K) 108 s (K + X) < 1026 GeV. (4.56)

For n = 2, we have M 6 & 1 TeV and the constraints are quite weak for n > 2.

2. Production of KK gravitons at high energy colliders: KK gravitons couple to the stress-

energy tensor T , so they can be attached anywhere in a process. The leading processes
for KK graviton production in high energy collisions are

e+ e /Z + GKK
q q g + GKK
qg q + GKK (4.57)

The Feynman diagrams for these processes are shown in Fig. 8, and the Feynman rules
can be found in Ref. [15, 16]. The experimental signals are missing energy/momentum
from the unobserved graviton [15, 16, 23]. The current bounds can be found in Ref. [24]
which is reproduced here.

Experiment and channel n=2 n=3 n=4 n=5 n=6

LEP Combined 1.60 1.20 0.94 0.77 0.66
CDF monophotons, 2.0 fb1 1.08 1.00 0.97 0.93 0.90
D monophotons, 2.7 fb1 0.97 0.90 0.87 0.85 0.83
CDF monojets, 1.1 fb1 1.31 1.08 0.98 0.91 0.88
CDF combined 1.42 1.16 1.06 0.99 0.95

18
4.2.3 Black hole productions at colliders and in cosmic rays

For Mpl,4+n TeV and ECM > Mpl,4+n , black holes can be produced at high energy colli-
sions [25, 26]. Semiclassical arguments work for MBH Mpl,4+n .
The Schwarzschild radius for a 4 + n-dimensional black hole is
 1
 n+1
1 MBH
Rs . (4.58)
Mpl,4+n Mpl,4+n

For 2 partons with s > Mpl,4+n moving in opposite directions, a black hole forms with mass

MBH s if the impact parameter is smaller than Rs . The cross section is given by the
geometrical formula,
  2
2 1 MBH n+2
(MBH ) Rs 2 , (4.59)
Mpl,4+n Mpl,4+n

which is large for s Mpl,4+n .
The produced black holes will decay through Hawking radiation with the Hawking tem-
perature
  1
Mpl,4+n n+1 1
TH Mpl,4+n . (4.60)
MBH Rs
A black hole decays equally to a particle on the brane or in the bulk, so it decays mostly to
the brane if there are more brane (SM) fields than the bulk fields (graviton). The multiplicity
of particles produced in a black hole evaporation can be estimated as
      n+2
MBH MBH MBH n+1
hNi = . (4.61)
Eparticle TH Mpl,4+n

The branching fraction to leptons is about 10% and to photons is about 2%. The search
strategy is to select events with high multiplicities and e or with E > 100 GeV. The reach
at the LHC extends up to Mpl,4+n 9 TeV [24]. High energy cosmic rays also provide powerful
probes of black hole productions in the large extra dimension scenario [27, 28, 29, 30].

4.3 Astrophysics bounds

Astrophysics provides some of the strongest constraints on the large extra dimension scenario
(for small n).

1. Star cooling: Constraints on the fundamental Planck scale can be obtained from bounds
on energy loss due to KK graviton emission from the Sun, red giants, and supernovae.

19
The rate for KK graviton emission can be estimated as
1 n Tn
2 (T R) n+2 . (4.62)
M4 M 4+n
We see that stronger bounds are obtained for higher temperature, and smaller n. The
temperatures for various stars are

Sun T keV ,
Red Giants T 100 keV ,
SN1987A T 30 MeV .

We will discuss the bounds from the supernova SN1987A as an example. The energy
loss of SN1987A is mostly due to neutrino emission. We can obtain a quick estimate
from the bound on the axion. The axion coupling to nucleons is mN /f where f
is the axion decay constant. On the other hand, the neutrino coupling is T2 /MW 2
.
12 3
Neutrinos are emitted from a neutrino sphere at R 15 km with 10 g/cm , T
4 MeV, beyond which there is no more scattering. The effective neutrino coupling is
T2
2
109 . (4.63)
MW
Requiring the axion coupling to be smaller than the neutrino coupling we obtain
mN
< 109 f & 109 GeV. (4.64)
f
2+n
Now comparing the KK graviton production cross section T n /M 4+n and the axion
production cross section 1/f 2 , we obtained the following dictionary for converting
experimental bounds,
1 Tn 1
2+n M 4+n & (T n flower
2
bound )
n+2 . (4.65)
f2 M 4n
For T 30 MeV, we have

M 4+n & 5 TeV, n = 2,

& 500 GeV, n = 3. (4.66)

More precise calculations lead to [31, 32, 33]

M6 & 14 TeV,
M7 & 1.6 TeV, (4.67)
n
where M4+n = (2) 2+n M 4+n .

20
 2. Diffuse ray background from long-lived KK graviton decays: If the KK gravitons
decay back to photons, it will affect the diffuse ray spectrum. The measurement of
EGRET satellite put bounds on the higher dimensional Planck scale M6 > 38 TeV,
M7 > 4.1 TeV [34, 35].

3. Most of the KK gravitons emitted by supernova remnants and neutron stars are gravita-
tionally trapped. The gravitons forming this halo occasionally decay, emitting photons.
Limits on rays from neutron star sources imply M6 > 200 TeV, M7 > 16 TeV [36].
The decay products of the gravitons forming the halo can hit the surface of the neutron
star, providing a heat source. The low measured luminosities of some pulsars implies
M6 > 750 TeV, M7 > 35 TeV [36].

The last 2 constraints assume how gravitons decay and hence have some model-dependence.
They may be evaded by some mild modifications of the theory (e.g., KK gravitons may decay
into hidden stuffs in some other branes).

4.4 Cosmological constraints

The earliest time we know about cosmology with some certainty is the era of Big-Bang
Nucleosynthesis (BBN) which starts at about T 1 MeV. In order to have BBN, the reheating
temperature after inflation must be higher than 1 MeV. On the other hand, high reheating
temperature means more KK gravitons are produced. They can affect

cooling of the universe due to graviton emission into extra dimensions,

expansion of the universe during BBN as KK gravitons redshift as R3 (radiation dom-

ination is required during BBN),

over-closure of the universe if graviton energy density is too large.

They put constraints on the reheating temperature as well as the higher-dimensional Planck
scale.

5 Physics in the Bulk

Large extra dimensions are an interesting proposal to address the hierarchy problem by bring-
ing down the fundamental Planck scale to TeV. However, this also removed the possibility
of using energy scales between the electroweak scale and the usual 4D Planck scale to address

21
 Symmetry Observable
breaking Mediator fields

Extra dimension

Figure 9: Suppression of symmetry breaking effects through extra dimensions.

other questions in particle physics, such as the fermion mass hierarchies. Indeed, one of the
major activities in particle physics is to explain small numbers (e.g., fermion masses, small-
ness or absence of flavor-changing effects, proton decay, . . . ). The traditional approach is to
employ symmetries. Extra dimensions provide some new ways to realize small numbers.

1. Large volume suppression: In addition to explain (M4+n /M4 )2 = (M4+n n

Vn )1 1,
the large volume in extra dimensions can be used to explain other small couplings.
For example, if right-handed neutrinos live in extra dimensions, their couplings to left-
handed neutrinos localized on the brane are suppressed by the large volume factor just
as the case for the gravitons. It can explain the smallness of the neutrino masses [37, 38].

2. Locality: Fields localized at different places in extra dimensions cannot couple directly.
Imaging that a coupling of SM fields is forbidden by some symmetry and the symmetry
is broken at a place far away from the SM in the extra dimensions. The symmetry
breaking effect is mediated to the SM sector by some mediator field . Then, the
induced coupling of the SM fields can be highly suppressed [39]:
1
Suppression factor , for rm 1 (massless mediator),
r n2
em r
, for rm 1 (massive mediator), (5.68)
r n2
where r is the distance between the SM brane and the source of the symmetry breaking
in extra dimensions.

22
 e e

KK, ZKK
e e

Figure 10: A four fermion interaction induced by KK gauge boson exchange.

6 Standard Model Fields in Extra Dimensions

So far we have considered gravity-only extra dimensions while the SM fields live on a 3-brane.
It is interesting to study the case that some of the SM fields also propagate in extra dimensions
(e.g., the 3-brane can have nonzero thickness). Because Standard Model has been tested up
to the TeV scale, the size of such extra dimensions must be . TeV1 . Since there may or
may not be other large extra dimensions where only gravity propagates, in discussing SM
fields propagating in extra dimensions we will take the fundamental Planck scale as a free
parameter and focus on the SM sector.
As SM fermions and Higgs carries gauge quantum numbers, they cannot propagate in
extra dimensions unless the corresponding gauge fields also propagate in extra dimensions.
On the other hand, SM fermions and Higgs may still be localized in 4 dimensions even if
gauge fields propagate in extra dimensions. As a result, there are several possibilities:

1. SM gauge fields propagate in extra dimensions while fermions and Higgs live on a
3-brane: In this case, higher-dimensional operators are induced by KK gauge boson
exchanges and they are strongly constrained by the electroweak precision data. For
example, Fig. 10 shows a 4-fermion operator induced by the KK photons and Z-bosons.
For one extra dimension, the size of the extra dimension is constrained to be R1 > 6.6
TeV [40].
Such a setup is useful for gaugino mediated SUSY breaking. If the SUSY is broken
at a place away from the SM fermions and their superpartners in extra dimensions
(Fig. 11), SUSY breaking is only transmitted to the MSSM sector by the gauge sector
which is flavor universal [41, 42, 43]. Unwanted flavor-non-universal interactions with
SUSY breaking sector are forbidden by locality in extra dimensions.

2. All SM fields live in the (same) extra dimensions: In this case they are called Universal
Extra Dimensions (UEDs) [44]. The experimental bound is weaker due to approximate

23
  
 
 

MSSM MSSM SUSY

matter gauge fields breaking

 
 
 

Figure 11: Gaugino mediated SUSY breaking through extra dimensions.

KK number conservation, R1 & 300 600 GeV [45, 46].

3. It is also possible to have the mixed scenario. Different SM fermions may even be
localized at different locations in extra dimensions. It can be used to explain the fermion
mass hierarchy as we will see next.

As mentioned earlier, gauge theories in more than 4 dimensions are non-renormalizable

and should be treated as low energy effective theories below some cutoff . The effective
expansion parameter is NKK4Nc . The theory becomes strongly coupled when NKK4Nc 1
or NKK 40, so one requires R . 40 for 1 extra dimension and R . 5 for 2 extra
dimensons. In the following subsections we discuss several interesting topics with SM living
in extra dimensions.

6.1 Split fermions fermion mass hierarchies without symmetries

from extra dimensions
Extra dimensions provide a new tool locality for doing physics in the bulk. In particular,
we can use it to explain the absence or smallness of certain interactions. For example, the
smallness of fermion Yukawa couplings and the absence of proton decays may be explained
if SM fermions are localized at different places in extra dimensions [47]. To see that, we first
consider how to localize a fermion in extra dimensions.
A chiral fermion can be localized in an extra dimension by a domain wall as shown in
Fig. 12. A fermion in 5 dimensions has 4 components. Its action is given by
Z
S = d4 xdy i + i5 5 + (y) ,
 
(6.69)

24


x5

Figure 12: A chiral fermion localized by a domain wall.

and the corresponding Dirac equation is


i + i5 5 + (y) = 0.

(6.70)

In the Weyl basis,

! ! !
0 I 0 L
= = , 5 = i , = , (6.71)
0 0 I R

where = (I, ~), = (I, ~ ). There are two Lorentz invariant fermion bilinears,
!
0
1 2 , T1 C5 2 , with C5 = 0 2 5 = . (6.72)
0

We look for solutions which are left- or right-handed 4D modes, i5 L = L , i5 R = R ,

(n) (n) (n) (n)
X X
(x, y) = fL (y)L (x) + fR (y)R (x), (6.73)
n n

(n) (n)
where L , R satisfy 4D Dirac equation with mass n .
Multiplying the 5D Dirac equation by the conjugate of the differential operator [i
(n) (n)
i5 5 + (y)] and using the 4D Dirac equation for L , R , we obtain
 (n)
i5 5 + (y) i 5 5 + (y) fL,R
 
 (n) (n)
= 52 + (y)2 (y) fL,R = 2n fL,R .

(6.74)

We can define creation and annihilation operators,

a = 5 + (y),
a = 5 + (y), (6.75)

25
then we have
 E  E
 (n)  (n)  (n)
a a fL = 52 + 2 fL = 2n fL ,

 E  E
 (n)  (n)  (n)
aa fR = 52 + 2 + fR = 2n fR .

(6.76)
(n) (n) (n) (n) (n) (n)
fL , fR each forms an orthonormal set. For 2n 6= 0, |fR i = 1n a|fL i, so L , R are
paired for nonzero modes.
Let us consider (y) = 22 y y, then the problem reduces to the simple harmonic

oscillator. Up to a factor of 2, a, a are just the annihilation and creation operators of
(n) (n)
a simple harmonic oscillator and the number operator N a a. The states |fL i, |fR i are
simply
(n) (n)
|fL i = |ni, |fR i = |n 1i. (6.77)
For the zero mode,
(0) (0)
a|fL i = 0 [5 + (y)]fL = 0, ,
(0) (0)
a |fR i = 0 [5 + (y)]fR = 0. (6.78)
The solutions are
Ry Ry
(0) (y )dy (0) (y )dy
fL (y) e 0 , fR (y) e 0 . (6.79)
(0)
Only one of them is normalizable for a domain wall: fL is normalizable if () < 0 and
(0)
(+) > 0, and fR is normalizable if () > 0 and (+) < 0. For (y) = 22 y,
(0) 1/2 2 y2
fL (y) = e . (6.80)
(/2)1/4
In this way we obtain a chiral fermion localized at y = 0.
One can generalize this setup to many fermion fields. Consider the action
Z X
S = d4 x dy [iM M + i (y) mi ]i . (6.81)
i

Each 5D fermion field i gives rise to a 4D chiral fermion. These chiral fermions are localized
around the zeros of i mi (Fig. 13). They can be used to explain the small couplings
from the small overlapping of wave functions in extra dimensions. We will consider a couple
examples below.
Yukawa couplings: Assuming that gauge and Higgs fields have flat wave functions in
the extra dimension, we consider the following action for the lepton fields,
Z
c
S = d5 xL[i 6 5 + (y)]L + E [i 6 5 + (y) m]E c

+(HLT C5 E c + h.c.) (6.82)

26
 1 (y)


2

3  m1 /1

 m2 /2
 m3 /3



Figure 13: Chiral fermions localized at difference places in extra dimensions by a domain wall.
r

l ec
h

Figure 14: Suppression of lepton Yukawa coupling by split fermions.

The zero mode of L field, l(0) is localized at y = 0 and the zero mode of E c field ec(0) is
m
localized at y = 22 r. The Yukawa coupling of the zero modes is then given by
Z Z
4 c (0) (0)
SYukawa = d x h(x)l(x)e (x) dyfl (y)fec (y), (6.83)

where
Z
2
Z
(0) (0) 2 2 2 (y m )2 2 r 2
dyfl (y)fec (y) = dy e y e 22 = e 2 . (6.84)

The Yukawa coupling is exponentially suppressed if the distance between the zero modes
r is somewhat bigger than 1 . (See Fig. 14)

Proton lifetime: The proton decay operator in 5D is

(QT C5 L) (U c T C5 D c )
Z
S d5 x
M3
(ql) (uc dc )
Z
d4 x , (6.85)
M2
where the suppression factor due to the separation of quark and lepton zero modes
(Fig. 15), r, is Z  2 2 3 2 2 3 2 2
dy e y e (yr) e 4 r . (6.86)

For r = 10, 1033 provides sufficient suppression for the proton decay rate.

27
 Quarks Leptons

thick wall

Figure 15: Suppression of proton decay by separation of quarks and leptons.

s s - y
0 R

Figure 16: An extra dimension on a line segment.

6.2 Boundary conditions and orbifolds

In Sec. 2 we discussed Kaluza-Klein theory for an extra dimensin compactified on a circle (or
a torus for more dimensions). To obtain realistic models we often need to compactify extra
dimensions on a line segment or an orbifold in order to have a chiral theory. In this section
we discuss such compactifications.
Consider a scalar field living in an extra dimension which is a line segment y [0, R]
(Fig. 16). The current in y directin is J5 = i 5 . Without addition sources at the bound-
aries we need J5 to vanish at the boundaries to preserve unitarity. Therefore, we should have
the following boundary conditions,

5 |boundary = 0 : Neumann boundary condition, or

|boundary = 0 : Dirichlet boundary condition. (6.87)

If we impose Neumann boundary conditions at both ends, 5 (y = 0) = 5 (y = R) = 0,

The KK decomposition of becomes
r
1 (0) 2 X (n) ny
+ (x, y) = + (x) + + (x) cos , (6.88)
R R n=1 R

with the mass mn = n/R for the nth mode. On the other hand , if we impose Dirichlet
boundary conditions at both ends, (y = 0) = (y = R) = 0, the KK decomposition of

28
 '$
r 6 6r r r
6

&%
???

Figure 17: The S 1 /Z2 orbifold.

is r
2 X (n) ny
(x, y) = (x) sin . (6.89)
R n=1 R
Note that there is no zero mode for Dirichlet boundary conditions.
The line segment can also be described as an orbifold S 1 /Z2 . If a theory has an exact
global symmetry, we can mod out (gauge) a subgroup of the global symmetry and still obtain
a consistent theory. For example, the torus compactification (S 1 ) which identifies (y + 2n)
and (y) is equivalent to gauging a discrete subgroup y y + 2n of translation in the y
direction. An orbifold is a space obtained by modding out a symmetry transformation of
another space which leaves some points fixed. The simplest example is S 1 /Z2. Starting with
a circle S 1 , we can gauge the Z2 symmetry y y (e.g., identifying points y and y), then
we obtain a line segment with fixed points y = 0 and y = R (Fig. 17).
Orbifold projection requires y y to be a good symmetry in the original theory, so
fields defined on [R, R] can be categorized as even or odd under y y. They are
equivalent to imposing Neumann or Dirichlet bondary conditions at the fixed points For a
scalar field,

even + : + (y) = + (y)

5 + (y = 0) = 5 + (y = R) = 0, Neumann BCs,
odd : (y) = (y)
(y = 0) = (y = R) = 0, Dirichlet BCs. (6.90)

We next consider a gauge field AM (x, y). Under the Z2 , x x , y y, so is even and
5 is odd. The field strength tensor F5 = A5 5 A needs to have a definite parity which
implies that A and A5 must have opposite parities. If A is even, then A5 is odd and has
no zero mode. Therefore in this orbifold compactification, there is no extra light scalar field

29
from the A5 component of the gauge field. Finally for the fermion field = (, )T ,
Z
S = d5 x iM M
Z
= d4 x dy (i + i + 5 5 ). (6.91)

Because 5 is odd, and must have opposite parities. Only one of them has a zero mode,
so the zero mode is chiral. This is also consistent with the gauge symmetry,

M AM A + A + A5 A5 . (6.92)

On can ask whether a mass term is allowed for the fermion if and have opposite parities,
m = m( + ). In the orbifold language m has to be odd. However, since the theory
is really just defined on one line segment, it is fine to include such a mass term. The mass
term will affect the localization of the fermion zero mode as it behaves like a domain wall.

6.3 Universal Extra Dimensions

Standard Model fermions are chiral. We see from the previous subsection how chiral fermions
arise in an orbifold compactification or from boundary conditions of extra dimensions. In
such a compactification, all SM fields can propagate in the same extra dimensions. This is
denoted as Universal Extra Dimensions (UEDs) [44].
Because all fields propagate in extra dimensions, the momentum in extra dimensions is
conserved except at the fixed points (boundaries). This translates to approximate KK number
conservation if the Lagrangian terms localized at the fixed points (boundaries) are ignored. As
a result, tree-level contributions to the electroweak observables from KK states are suppressed.
They can contribute at one-loop level. The bound on the size of UEDs is only 1/R & 300600
GeV [45, 46].
However, boundary terms will be present as they are not forbidden by symmetries and they
are induced by bulk loop corrections [48, 49]. The boundary terms modify the KK spectrum
from m(n) = n/R and lift the degeneracies of KK excitations of different SM species. Some
discrete subgroup of KK number can still be preserved even including radiative corrections.
For example in S 1 /Z2 compactification, a Z2 reflection about y = R/2 is a good symmetry if
the boundary terms at y = 0 and y = R are equal. (which is the case from bulk loops). As
a result, the odd KK levels are odd and even KK levels are even under an exact KK-parity.
It implies that the first (odd) KK-level states have to be pair produced and the lightest first
KK excitation of all SM fields is stable. It can be a good dark matter candidate if it is

30
neutral [50, 51]. For collider phenomenology, the first KK states are pair-produced, then each
goes through cascade decays which ends up with the stable lightest KK state. The signatures
are jets/leptons plus missing energy, similar to those of SUSY with conserved R-parity [52].
More detailed measurements are required to distinguish these theories.

6.4 Symmetry breaking by orbifolds (boundary conditions)

When there is a symmetry of the Lagrangian, the fields at the identified points in a com-
pactification do not have to be identical, but merely equal up to a symmetry transformation.
In the orbifold language, instead of modding out a discrete subgroup of the space-time sym-
metry, we can mod out a diagonal combination of the space-time symmetry and the internal
symmetry. In this way, we can achieve symmetry breaking by orbifold compactifications.
As an example, let us consider the breaking of the grand unified gauge group SU(5) to
the SM gauge group SU(3) SU(2) U(1) on S 1 /Z2 orbifold. The embedding of SU(3)
SU(2) U(1) in SU(5) is

2
SU(3) X, Y 2



SU(5) =
, U(1)
2 .
(6.93)
3


X, Y SU(2) 3

Under the Z2 subgroup of SU(5), exp[i diag(2, 2, 2, 3, 3)], the fundamental representation
5 transforms as
+
+


5:+ ,
(6.94)



and the adjoint representation 24 transforms as

+ + +
+ + +



24 :
+ + + .
(6.95)
+ +


+ +

31
Choosing the Z2 of S 1 /Z2 orbifold to be the diagonal combination of the above Z2 subgroup
of SU(5) and y y, then the transformation properties of various gauge field components
are
SU(3) SU(2) U(1) X, Y
A +
A5 +

We see that only 4D gauge fields corresponding to SU(3) SU(2) U(1) have zero modes,
so SU(5) is broken down to SU(3) SU(2) U(1) [53, 54, 55, 56].
This can be described equivalently by the following boundary conditions,

SU(3) SU(2) U(1) X, Y

A (N, N) (D, D)
A5 (D, D) (N, N)

where N and D represent Neumann and Dirichelet boundary conditions respectively with first
entry referring to y = 0 and second entry referring to y = R. In other words, while the bulk
has the SU(5) gauge symmetry, the boundaries only preserve SU(3) SU(2) U(1).
For these boundary conditions, the A5 components of the off-diagonal gauge bosons X, Y
have zero modes. They appear as light charged scalars in this theory. Their masses are
g 1
generated from finite one-loop corrections and are of order m 4 R
. These light scalars can
be removed by choosing a different set of boundary conditions,

SU(3) SU(2) U(1) X, Y

A (N, N) (N, D)
A5 (D, D) (D, N)

In this setup, the bulk and one of the boundaries (y = 0) preserve the full SU(5) gauge
symmetry, the other boundary (y = R) only preserves SU(3) SU(2) U(1). In the
orbifold language, this corresponds to an S 1 /(Z2 Z2 ) orbifold.
We have seen that an orbifold compactification can be equivalently described by boundary
conditions. In fact, we can also consider more general boundary conditions [56, 57]. Let us
start with a bulk action for the scalar field
Z R  
1 M
Z
4
Sbulk = d x dy M V () , (6.96)
0 2

32
Consider first for simplicity that there is no boundary term. Applying the variation principle
we obtain
Z R  
V
Z
4 M
S = d x dy M
0
Z R  
V
Z
4
= d x dy y y
0
Z R   Z R
V
Z
4 M 4
= d x dy M d xy . (6.97)
0 0

For S = 0 we require that the field satisfies the bulk equation of motion (EOM)
V
M M = (6.98)

and also the boundary variation needs to vanish,

y |boundary = 0. (6.99)

This can be satisfied by the Neumann boundary condition y |boundary = 0 for arbitrary
|boundary . We call it natural boundary condition if |boundary is left free to vary.
Other boundary conditions can be obtained by adding boundary terms. For example, we
can include boundary mass terms,

4 1 1
Z Z
S = Sbulk d x M1 |y=0 d4 x M22 2 |y=R .
2 2
(6.100)
2 2
Variation of the action gives
Z
Sboundary = d5 x (EOM)
Z Z
(y + M2 )|y=R + d4 x(y M12 )|y=0 .
2
(6.101)

The natural BCs will be given by

y + M22 = 0 at y = R,
y M12 = 0 at y = 0. (6.102)

If we take M1 , M2 , then = 0 at y = 0, R and y will be arbitrary at y = , R

, infinitesimally away from the boundaries. This is equivalent to the Dirichlet boundary
conditions. We can always understand the Dirichlet boundary conditions as the case with
infinitely large boundary mass terms for the fields.

33
 Similarly for a gauge field,
1 a 1 a a 1 a 5 a
Z Z
5 MN a
S = d x( FM N F ) = d5 x( F F F5 F ), (6.103)
4 4 2
the natural boundary conditions in unitary gauge are y Aa = 0, Aa5 = 0.
Now we add boundary scalar fields which have nonzero VEVs,
1
Li = |D i |2 i (|i |2 vi2 )2 , (6.104)
2
where i = 1 is at y = 0 and i = 2 is at y = R. These boundary terms will induce
non-vanishing VEVs and i can be parametrized as a physical Higgs and a Goldstone boson,
1
i = (vi + hi )eii /vi . (6.105)
2
In this case, the natural boundary conditions become
2
y A v1,2 A |1,2 = 0. (6.106)

Taking vi , we obtain A |1,2 = 0, which corresponds to Dirichlet bondary conditions. The

Higgs and the Goldstone hi , i decouple from the gauge field in this limit. At the same time,
A5 boundary conditions change from Dirichlet to Neumann boundary conditions y A5 = 0.
Note that the mass of the lightest gauge boson in this limit is 1/R, independent of vi .
Gauge symmetry breaking can be achieved by choosing appropriate boundary conditions.
For SU(2) U(1), we can set A1,2 1,2 3 3
(y = 0) = A (y = R) = 0, y A (y = 0) = y A (y =
R) = 0. This is equivalent to orbifold breaking. However, we can also reduce the rank of the
gauge group by breaking SU(2) to nothing, which can not be achieved by a simple orbifold.
We just need to choose A1,2 3 2,3 1
(y = 0) = 0, y A (y = 0) = 0, and A (y = R) = 0, y A (y =
R) = 0, then no gauge field has a zero mode and the gauge symmetry is completely broken.
The discussion of boundary conditions for fermions can be found in Ref. [58]. One can
build realistic models with electroweak symmetry broken by boundary conditions [58, 59].

7 Dimension Deconstruction
Gauge theories in more than 4 dimensions are non-renormalizable. They are treated as
low energy effective theories below some cutoff . A nave momentum cutoff breaks gauge
invariance. It also violates locality in extra dimensions. We need to be careful about how to
implement the cutoff and asking cutoff-sensitive questions, otherwise we can easily get non-
sensible answers if we do not regularize the theories in a correct way. One way to regularize

34
 '$'$'$ '$
1 2 ...
SU(Nc )0 SU(Nc )1 SU(Nc )2 SU(Nc )N
&%&%&% &%

Figure 18: A lattice of SU(Nc ) gauge groups.

the higher dimensional gauge theories while preserving gauge invariance and locality is to put
extra dimensions on a lattice [60, 61, 62].
Consider N + 1 copies of SU(Nc ) gauge groups in 4D, with N link-Higgs fields i which
transform as bi-fundamentals (Nc,i , N c,i1 ) under neighboring gauge groups (Fig. 18),
N N
1 X a ia X
LQCD = F F + D i D i , (7.107)
4 i=0 i i=1

N
X
D = + ig Aai Tia , (7.108)
i=0

where g is the gauge coupling of SU(Nc )i group (assumed to be identical for simplicity). We
can write down a potential for each link-Higgs field such that each i develops a VEV of the
form
hi i = v I (7.109)
where I is the (N + 1) (N + 1) identity matrix. We assume that all left-over physical Higgs
bosons are heavy (with mass > v).
These VEVs break N +1 SU(Nc ) gauge groups down to the diagonal SU(Nc ) gauge group.
The mass matrix for the N + 1 sets of gauge fields Aai , (i = 0, , N) is

1 1 0 0
1 2 1 0


1 2 2 0 1 2 0

M = g v . .. . (7.110)
2 .. .

0 0 2 1

0 0 1 1

The mass eigenstates An can be obtained by diagonalizing the mass matrix,

N
X
Aj = ajn An . (7.111)
n=0

35
For n 6= 0, r
2 2j + 1
ajn = cos ( n ) , j = 0, 1, . . . , N, (7.112)
N +1 2
where n = n/(N + 1), and
1
aj0 = j = 0, 1, . . . , N. (7.113)
N +1

The masses for An are given by

   
n n
Mn = 2gv sin = 2gv sin , n = 0, 1, . . . , N. (7.114)
2 2(N + 1)
For small n,
gvn
Mn , n N. (7.115)
(N + 1)
They are the same as the KK tower from a compactified extra dimension if we identify
gv 1 N +1
= = L= . (7.116)
(N + 1) R L gv

This corresponds to compactification on a line segment (or S 1 /Z2 orbifold). The low energy
gauge coupling of the remaining diagonal gauge group is given by

g = g/ N + 1. (7.117)

The 4D theory with N + 1 SU(Nc ) gauge groups provides a renormalizable UV completion

of 5D gauge theory with a cutoff
1 1
= , (7.118)
a gv
where a is the lattice spacing. Many results in higher-dimensional theories can be easily
translated into the 4D language with this dimension-deconstruction. For example, the orbifold
symmetry breaking of SU(5) SU(3) SU(2) U(1) can be realized by considering many
copies of SU(5) gauge groups with one copy of SU(3) SU(2) U(1) gauge group at the
end in 4D, broken by the link fields to the diagonal SU(3) SU(2) U(1) gauge group
(Fig. 19) [63, 64].
For gauge theories, there is one-to-one correspondence between the higher-dimensional
theory and a 4D theory with a theory space. However, dimension-deconstruction for gravity
is more difficult, there are strong-coupling issues involving the scalar longitudinal degrees of
freedom and the cutoff is much lower than the navely expected value [65]. It is beyond the
scope of this lecture. Interested readers are referred to the original paper of this discussion.

36
 '$'$'$ '$
1 2 SU(3)
SU(5) SU(5) SU(5) ... SU(2)
&%&%&% &%
U(1)

Figure 19: The deconstructed version of orbifold symmetry breaking.

8 Epilogue
In these lectures I tried to give a brief summary of theories with flat extra dimensions de-
veloped in recent years. The goal is to give a jump start for advanced graduate students
who are interested in research in this area. The topics covered here are certainly not com-
plete, but just a sample of what people have been thinking of in these directions. Hope-
fully they provide enough basics so that the students are able to find and understand the
other related subjects in the literature. There are also many other useful review articles and
lectures on extra dimensions with different emphases which are good resources for further
explorations [66, 67, 68, 69, 70, 71, 72].

Acknowledgments
I would like to thank Csaba Csaki for organizing the TASI 2009 summer school and inviting
me to give the lectures. This work is supported in part by the Department of Energy Grant
DE-FG02-91ER40674.

References
[1] Nordstrom, G., Physikalische Zeitschrift 15: 504506, OCLC 1762351 (1914).

[2] Kaluza, T., Sitzungsber. Preuss. Akad. Wiss. Berlin (Math. Phys.) 1921: 966972 (1921).

[3] Klein, O., Zeitschrift fr Physik a Hadrons and Nuclei 37 (12): 895906
doi:10.1007/BF01397481 (1926).

[4] I. Antoniadis, Phys. Lett. B 246, 377 (1990).

[5] P. Horava and E. Witten, Nucl. Phys. B 460, 506 (1996) [arXiv:hep-th/9510209].

[6] P. Horava and E. Witten, Nucl. Phys. B 475, 94 (1996) [arXiv:hep-th/9603142].

37
 [7] J. Polchinski, Phys. Rev. Lett. 75, 4724 (1995) [arXiv:hep-th/9510017].

[8] N. Arkani-Hamed, S. Dimopoulos and G. R. Dvali, Phys. Lett. B 429, 263 (1998)
[arXiv:hep-ph/9803315].

[9] L. Randall and R. Sundrum, Phys. Rev. Lett. 83, 3370 (1999) [arXiv:hep-ph/9905221].

[10] L. Randall and R. Sundrum, Phys. Rev. Lett. 83, 4690 (1999) [arXiv:hep-th/9906064].

[11] J. M. Maldacena, Adv. Theor. Math. Phys. 2, 231 (1998) [Int. J. Theor. Phys. 38, 1113
(1999)] [arXiv:hep-th/9711200].

[12] T. Gherghetta, lectures at TASI 2009.

[13] I. Antoniadis, N. Arkani-Hamed, S. Dimopoulos and G. R. Dvali, Phys. Lett. B 436, 257
(1998) [arXiv:hep-ph/9804398].

[14] N. Arkani-Hamed, S. Dimopoulos and G. R. Dvali, Phys. Rev. D 59, 086004 (1999)
[arXiv:hep-ph/9807344].

[15] G. F. Giudice, R. Rattazzi and J. D. Wells, Nucl. Phys. B 544, 3 (1999)

[arXiv:hep-ph/9811291].

[16] T. Han, J. D. Lykken and R. J. Zhang, Phys. Rev. D 59, 105006 (1999)
[arXiv:hep-ph/9811350].

[17] R. Sundrum, Phys. Rev. D 59, 085009 (1999) [arXiv:hep-ph/9805471].

[18] D. J. Kapner, T. S. Cook, E. G. Adelberger, J. H. Gundlach, B. R. Heckel, C. D. Hoyle

and H. E. Swanson, Phys. Rev. Lett. 98, 021101 (2007) [arXiv:hep-ph/0611184].

[19] W. Skiba, lectures at TASI 2009.

[20] S. Nussinov and R. Shrock, Phys. Rev. D 59, 105002 (1999) [arXiv:hep-ph/9811323].

[21] J. L. Hewett, Phys. Rev. Lett. 82, 4765 (1999) [arXiv:hep-ph/9811356].

[22] M. L. Graesser, Phys. Rev. D 61, 074019 (2000) [arXiv:hep-ph/9902310].

[23] E. A. Mirabelli, M. Perelstein and M. E. Peskin, Phys. Rev. Lett. 82, 2236 (1999)
[arXiv:hep-ph/9811337].

[24] G. Landsberg, arXiv:0808.1867 [hep-ex].

38
[25] S. Dimopoulos and G. L. Landsberg, Phys. Rev. Lett. 87, 161602 (2001)
[arXiv:hep-ph/0106295].

[26] S. B. Giddings and S. D. Thomas, Phys. Rev. D 65, 056010 (2002)

[arXiv:hep-ph/0106219].

[27] J. L. Feng and A. D. Shapere, Phys. Rev. Lett. 88, 021303 (2002)
[arXiv:hep-ph/0109106].

[28] L. Anchordoqui and H. Goldberg, Phys. Rev. D 65, 047502 (2002)

[arXiv:hep-ph/0109242].

[29] R. Emparan, M. Masip and R. Rattazzi, Phys. Rev. D 65, 064023 (2002)
[arXiv:hep-ph/0109287].

[30] L. A. Anchordoqui, J. L. Feng, H. Goldberg and A. D. Shapere, Phys. Rev. D 65, 124027
(2002) [arXiv:hep-ph/0112247].

[31] S. Cullen and M. Perelstein, Phys. Rev. Lett. 83, 268 (1999) [arXiv:hep-ph/9903422].

[32] V. D. Barger, T. Han, C. Kao and R. J. Zhang, Phys. Lett. B 461, 34 (1999)
[arXiv:hep-ph/9905474].

[33] C. Hanhart, J. A. Pons, D. R. Phillips and S. Reddy, Phys. Lett. B 509, 1 (2001)
[arXiv:astro-ph/0102063].

[34] L. J. Hall and D. Tucker-Smith, Phys. Rev. D 60, 085008 (1999) [arXiv:hep-ph/9904267].

[35] S. Hannestad and G. Raffelt, Phys. Rev. Lett. 87, 051301 (2001) [arXiv:hep-ph/0103201].

[36] S. Hannestad and G. G. Raffelt, Phys. Rev. Lett. 88, 071301 (2002)
[arXiv:hep-ph/0110067].

[37] K. R. Dienes, E. Dudas and T. Gherghetta, Nucl. Phys. B 557, 25 (1999)

[arXiv:hep-ph/9811428].

[38] N. Arkani-Hamed, S. Dimopoulos, G. R. Dvali and J. March-Russell, Phys. Rev. D 65,

024032 (2002) [arXiv:hep-ph/9811448].

[39] N. Arkani-Hamed and S. Dimopoulos, Phys. Rev. D 65, 052003 (2002)

[arXiv:hep-ph/9811353].

39
[40] K. m. Cheung and G. L. Landsberg, Phys. Rev. D 65, 076003 (2002)
[arXiv:hep-ph/0110346].

[41] E. A. Mirabelli and M. E. Peskin, Phys. Rev. D 58, 065002 (1998)

[arXiv:hep-th/9712214].

[42] D. E. Kaplan, G. D. Kribs and M. Schmaltz, Phys. Rev. D 62, 035010 (2000)
[arXiv:hep-ph/9911293].

[43] Z. Chacko, M. A. Luty, A. E. Nelson and E. Ponton, JHEP 0001, 003 (2000)
[arXiv:hep-ph/9911323].

[44] T. Appelquist, H. C. Cheng and B. A. Dobrescu, Phys. Rev. D 64, 035002 (2001)
[arXiv:hep-ph/0012100].

[45] T. Appelquist and H. U. Yee, Phys. Rev. D 67, 055002 (2003) [arXiv:hep-ph/0211023].

[46] I. Gogoladze and C. Macesanu, Phys. Rev. D 74, 093012 (2006) [arXiv:hep-ph/0605207].

[47] N. Arkani-Hamed and M. Schmaltz, Phys. Rev. D 61, 033005 (2000)

[arXiv:hep-ph/9903417].

[48] H. Georgi, A. K. Grant and G. Hailu, Phys. Lett. B 506, 207 (2001)
[arXiv:hep-ph/0012379].

[49] H. C. Cheng, K. T. Matchev and M. Schmaltz, Phys. Rev. D 66, 036005 (2002)
[arXiv:hep-ph/0204342].

[50] G. Servant and T. M. P. Tait, Nucl. Phys. B 650, 391 (2003) [arXiv:hep-ph/0206071].

[51] H. C. Cheng, J. L. Feng and K. T. Matchev, Phys. Rev. Lett. 89, 211301 (2002)
[arXiv:hep-ph/0207125].

[52] H. C. Cheng, K. T. Matchev and M. Schmaltz, Phys. Rev. D 66, 056006 (2002)
[arXiv:hep-ph/0205314].

[53] Y. Kawamura, Prog. Theor. Phys. 105, 999 (2001) [arXiv:hep-ph/0012125].

[54] G. Altarelli and F. Feruglio, Phys. Lett. B 511, 257 (2001) [arXiv:hep-ph/0102301].

[55] L. J. Hall and Y. Nomura, Phys. Rev. D 64, 055003 (2001) [arXiv:hep-ph/0103125].

40
[56] A. Hebecker and J. March-Russell, Nucl. Phys. B 625, 128 (2002)
[arXiv:hep-ph/0107039].

[57] C. Csaki, C. Grojean, H. Murayama, L. Pilo and J. Terning, Phys. Rev. D 69, 055006
(2004) [arXiv:hep-ph/0305237].

[58] C. Csaki, C. Grojean, J. Hubisz, Y. Shirman and J. Terning, Phys. Rev. D 70, 015012
(2004) [arXiv:hep-ph/0310355].

[59] C. Csaki, C. Grojean, L. Pilo and J. Terning, Phys. Rev. Lett. 92, 101802 (2004)
[arXiv:hep-ph/0308038].

[60] N. Arkani-Hamed, A. G. Cohen and H. Georgi, Phys. Rev. Lett. 86, 4757 (2001)
[arXiv:hep-th/0104005].

[61] C. T. Hill, S. Pokorski and J. Wang, Phys. Rev. D 64, 105005 (2001)
[arXiv:hep-th/0104035].

[62] H. C. Cheng, C. T. Hill, S. Pokorski and J. Wang, Phys. Rev. D 64, 065007 (2001)
[arXiv:hep-th/0104179].

[63] C. Csaki, G. D. Kribs and J. Terning, Phys. Rev. D 65, 015004 (2002)
[arXiv:hep-ph/0107266].

[64] H. C. Cheng, K. T. Matchev and J. Wang, Phys. Lett. B 521, 308 (2001)
[arXiv:hep-ph/0107268].

[65] N. Arkani-Hamed, H. Georgi and M. D. Schwartz, Annals Phys. 305, 96 (2003)

[arXiv:hep-th/0210184].

[66] V. A. Rubakov, Phys. Usp. 44, 871 (2001) [Usp. Fiz. Nauk 171, 913 (2001)]
[arXiv:hep-ph/0104152].

[67] G. Gabadadze, arXiv:hep-ph/0308112.

[68] C. Csaki, arXiv:hep-ph/0404096.

[69] A. Perez-Lorenzana, J. Phys. Conf. Ser. 18, 224 (2005) [arXiv:hep-ph/0503177].

[70] R. Sundrum, arXiv:hep-th/0508134.

[71] C. Csaki, J. Hubisz and P. Meade, arXiv:hep-ph/0510275.

41
[72] R. Rattazzi, arXiv:hep-ph/0607055.

42